Rewrote uniq to be less than a third of the size, and fixed some other

minor problems.

diff --git a/busybox.h b/busybox.h
index 69f45543576e0c28063ad9fdd671a333d3fd51fb..faad206ee1da400e7d8bd694e80e26150cce409c 100644 (file)
--- a/busybox.h
+++ b/busybox.h
@@ -395,6 +395,7 @@ extern int print_file_by_name(char *filename);
 extern char process_escape_sequence(char **ptr);
 extern char *get_last_path_component(char *path);
 extern void xregcomp(regex_t *preg, const char *regex, int cflags);
+extern FILE *xfopen(const char *path, const char *mode);
 
 #ifndef DMALLOC 
 extern void *xmalloc (size_t size);

diff --git a/coreutils/uniq.c b/coreutils/uniq.c
index ef38587bd2066479de1e703bcd63c2cfdf91151d..cfe6cca5e2d23026b6766c824e8b24d37742cc70 100644 (file)
--- a/coreutils/uniq.c
+++ b/coreutils/uniq.c
@@ -5,6 +5,7 @@
  *
  * Copyright (C) 1999,2000 by Lineo, inc.
  * Written by John Beppu <beppu@lineo.com>
+ * Rewritten by Matt Kraai <kraai@alumni.carnegiemellon.edu>
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -27,156 +28,29 @@
 #include <string.h>
 #include <errno.h>
 
-/* max chars in line */
-#define UNIQ_MAX 4096
-
-typedef void (Print) (FILE *, const char *);
-
-typedef int (Decide) (const char *, const char *);
-
-/* container for two lines to be compared */
-typedef struct {
-       char *a;
-       char *b;
-       int recurrence;
-       FILE *in;
-       FILE *out;
-       void *func;
-} Subject;
-
-/* set up all the variables of a uniq operation */
-static Subject *subject_init(Subject * self, FILE * in, FILE * out,
-                                                        void *func)
-{
-       self->a = NULL;
-       self->b = NULL;
-       self->in = in;
-       self->out = out;
-       self->func = func;
-       self->recurrence = 0;
-       return self;
-}
-
-/* point a and b to the appropriate lines;
- * count the recurrences (if any) of a string;
- */
-static Subject *subject_next(Subject * self)
-{
-       /* tmp line holders */
-       static char line[2][UNIQ_MAX];
-       static int alternator = 0;
-
-       if (fgets(line[alternator], UNIQ_MAX, self->in)) {
-               self->a = self->b;
-               self->b = line[alternator];
-               alternator ^= 1;
-               return self;
-       }
-
-       return NULL;
-}
-
-static Subject *subject_last(Subject * self)
-{
-       self->a = self->b;
-       self->b = NULL;
-       return self;
-}
-
-static Subject *subject_study(Subject * self)
-{
-       if (self->a == NULL) {
-               return self;
-       }
-       if (self->b == NULL) {
-               fprintf(self->out, "%s", self->a);
-               return self;
-       }
-       if (strcmp(self->a, self->b) == 0) {
-               self->recurrence++;
-       } else {
-               fprintf(self->out, "%s", self->a);
-               self->recurrence = 0;
-       }
-       return self;
-}
-
-static int
-set_file_pointers(int schema, FILE ** in, FILE ** out, char **argv)
-{
-       switch (schema) {
-       case 0:
-               *in = stdin;
-               *out = stdout;
-               break;
-       case 1:
-               *in = fopen(argv[0], "r");
-               *out = stdout;
-               break;
-       case 2:
-               *in = fopen(argv[0], "r");
-               *out = fopen(argv[1], "w");
-               break;
-       }
-       if (*in == NULL) {
-               errorMsg("%s: %s\n", argv[0], strerror(errno));
-               return errno;
-       }
-       if (*out == NULL) {
-               errorMsg("%s: %s\n", argv[1], strerror(errno));
-               return errno;
-       }
-       return 0;
-}
-
-
-/* one variable is the decision algo */
-/* another variable is the printing algo */
-
-/* I don't think I have to have more than a 1 line memory 
-   this is the one constant */
-
-/* it seems like GNU/uniq only takes one or two files as an option */
-
-/* ________________________________________________________________________ */
 int uniq_main(int argc, char **argv)
 {
-       int i;
-       char opt;
-       FILE *in, *out;
-       Subject s;
+       FILE *in = stdin, *out = stdout;
+       char *lastline = NULL, *input;
 
        /* parse argv[] */
-       for (i = 1; i < argc; i++) {
-               if (argv[i][0] == '-') {
-                       opt = argv[i][1];
-                       switch (opt) {
-                       case '-':
-                       case 'h':
-                               usage(uniq_usage);
-                       default:
-                               usage(uniq_usage);
-                       }
-               } else {
-                       break;
-               }
-       }
+       if ((argc > 1 && **(argv + 1) == '-') || argc > 3)
+               usage(uniq_usage);
 
-       /* 0 src: stdin; dst: stdout */
-       /* 1 src: file;  dst: stdout */
-       /* 2 src: file;  dst: file   */
-       if (set_file_pointers((argc - 1), &in, &out, &argv[i])) {
-               exit(1);
+       if (argv[1] != NULL) {
+               in = xfopen(argv[1], "r");
+               if (argv[2] != NULL)
+                       out = xfopen(argv[2], "w");
        }
 
-       subject_init(&s, in, out, NULL);
-       while (subject_next(&s)) {
-               subject_study(&s);
+       while ((input = get_line_from_file(in)) != NULL) {
+               if (lastline == NULL || strcmp(input, lastline) != 0) {
+                       fputs(input, out);
+                       free(lastline);
+                       lastline = input;
+               }
        }
-       subject_last(&s);
-       subject_study(&s);
+       free(lastline);
 
-       return(0);
+       return EXIT_SUCCESS;
 }
-
-/* $Id: uniq.c,v 1.14 2000/09/25 21:45:58 andersen Exp $ */

diff --git a/include/busybox.h b/include/busybox.h
index 69f45543576e0c28063ad9fdd671a333d3fd51fb..faad206ee1da400e7d8bd694e80e26150cce409c 100644 (file)
--- a/include/busybox.h
+++ b/include/busybox.h
@@ -395,6 +395,7 @@ extern int print_file_by_name(char *filename);
 extern char process_escape_sequence(char **ptr);
 extern char *get_last_path_component(char *path);
 extern void xregcomp(regex_t *preg, const char *regex, int cflags);
+extern FILE *xfopen(const char *path, const char *mode);
 
 #ifndef DMALLOC 
 extern void *xmalloc (size_t size);

diff --git a/uniq.c b/uniq.c
index ef38587bd2066479de1e703bcd63c2cfdf91151d..cfe6cca5e2d23026b6766c824e8b24d37742cc70 100644 (file)
--- a/uniq.c
+++ b/uniq.c
@@ -5,6 +5,7 @@
  *
  * Copyright (C) 1999,2000 by Lineo, inc.
  * Written by John Beppu <beppu@lineo.com>
+ * Rewritten by Matt Kraai <kraai@alumni.carnegiemellon.edu>
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -27,156 +28,29 @@
 #include <string.h>
 #include <errno.h>
 
-/* max chars in line */
-#define UNIQ_MAX 4096
-
-typedef void (Print) (FILE *, const char *);
-
-typedef int (Decide) (const char *, const char *);
-
-/* container for two lines to be compared */
-typedef struct {
-       char *a;
-       char *b;
-       int recurrence;
-       FILE *in;
-       FILE *out;
-       void *func;
-} Subject;
-
-/* set up all the variables of a uniq operation */
-static Subject *subject_init(Subject * self, FILE * in, FILE * out,
-                                                        void *func)
-{
-       self->a = NULL;
-       self->b = NULL;
-       self->in = in;
-       self->out = out;
-       self->func = func;
-       self->recurrence = 0;
-       return self;
-}
-
-/* point a and b to the appropriate lines;
- * count the recurrences (if any) of a string;
- */
-static Subject *subject_next(Subject * self)
-{
-       /* tmp line holders */
-       static char line[2][UNIQ_MAX];
-       static int alternator = 0;
-
-       if (fgets(line[alternator], UNIQ_MAX, self->in)) {
-               self->a = self->b;
-               self->b = line[alternator];
-               alternator ^= 1;
-               return self;
-       }
-
-       return NULL;
-}
-
-static Subject *subject_last(Subject * self)
-{
-       self->a = self->b;
-       self->b = NULL;
-       return self;
-}
-
-static Subject *subject_study(Subject * self)
-{
-       if (self->a == NULL) {
-               return self;
-       }
-       if (self->b == NULL) {
-               fprintf(self->out, "%s", self->a);
-               return self;
-       }
-       if (strcmp(self->a, self->b) == 0) {
-               self->recurrence++;
-       } else {
-               fprintf(self->out, "%s", self->a);
-               self->recurrence = 0;
-       }
-       return self;
-}
-
-static int
-set_file_pointers(int schema, FILE ** in, FILE ** out, char **argv)
-{
-       switch (schema) {
-       case 0:
-               *in = stdin;
-               *out = stdout;
-               break;
-       case 1:
-               *in = fopen(argv[0], "r");
-               *out = stdout;
-               break;
-       case 2:
-               *in = fopen(argv[0], "r");
-               *out = fopen(argv[1], "w");
-               break;
-       }
-       if (*in == NULL) {
-               errorMsg("%s: %s\n", argv[0], strerror(errno));
-               return errno;
-       }
-       if (*out == NULL) {
-               errorMsg("%s: %s\n", argv[1], strerror(errno));
-               return errno;
-       }
-       return 0;
-}
-
-
-/* one variable is the decision algo */
-/* another variable is the printing algo */
-
-/* I don't think I have to have more than a 1 line memory 
-   this is the one constant */
-
-/* it seems like GNU/uniq only takes one or two files as an option */
-
-/* ________________________________________________________________________ */
 int uniq_main(int argc, char **argv)
 {
-       int i;
-       char opt;
-       FILE *in, *out;
-       Subject s;
+       FILE *in = stdin, *out = stdout;
+       char *lastline = NULL, *input;
 
        /* parse argv[] */
-       for (i = 1; i < argc; i++) {
-               if (argv[i][0] == '-') {
-                       opt = argv[i][1];
-                       switch (opt) {
-                       case '-':
-                       case 'h':
-                               usage(uniq_usage);
-                       default:
-                               usage(uniq_usage);
-                       }
-               } else {
-                       break;
-               }
-       }
+       if ((argc > 1 && **(argv + 1) == '-') || argc > 3)
+               usage(uniq_usage);
 
-       /* 0 src: stdin; dst: stdout */
-       /* 1 src: file;  dst: stdout */
-       /* 2 src: file;  dst: file   */
-       if (set_file_pointers((argc - 1), &in, &out, &argv[i])) {
-               exit(1);
+       if (argv[1] != NULL) {
+               in = xfopen(argv[1], "r");
+               if (argv[2] != NULL)
+                       out = xfopen(argv[2], "w");
        }
 
-       subject_init(&s, in, out, NULL);
-       while (subject_next(&s)) {
-               subject_study(&s);
+       while ((input = get_line_from_file(in)) != NULL) {
+               if (lastline == NULL || strcmp(input, lastline) != 0) {
+                       fputs(input, out);
+                       free(lastline);
+                       lastline = input;
+               }
        }
-       subject_last(&s);
-       subject_study(&s);
+       free(lastline);
 
-       return(0);
+       return EXIT_SUCCESS;
 }
-
-/* $Id: uniq.c,v 1.14 2000/09/25 21:45:58 andersen Exp $ */

diff --git a/utility.c b/utility.c
index f7dda7917d08db50c169f76ec041a623597efb0b..d376a04fb5382bcdd36d96d6798708bd64e9ff84 100644 (file)
--- a/utility.c
+++ b/utility.c
@@ -81,7 +81,7 @@ extern void usage(const char *usage)
 {
        fprintf(stderr, "%s\n\n", full_version);
        fprintf(stderr, "Usage: %s\n", usage);
-       exit FALSE;
+       exit(EXIT_FAILURE);
 }
 
 extern void errorMsg(const char *s, ...)
@@ -106,7 +106,7 @@ extern void fatalError(const char *s, ...)
        vfprintf(stderr, s, p);
        va_end(p);
        fflush(stderr);
-       exit( FALSE);
+       exit(EXIT_FAILURE);
 }
 
 #if defined BB_INIT
@@ -401,17 +401,17 @@ copyFile(const char *srcName, const char *destName,
                /* This is fine, since symlinks never get here */
                if (chown(destName, srcStatBuf.st_uid, srcStatBuf.st_gid) < 0) {
                        perror(destName);
-                       exit FALSE;
+                       exit(EXIT_FAILURE);
                }
                if (chmod(destName, srcStatBuf.st_mode) < 0) {
                        perror(destName);
-                       exit FALSE;
+                       exit(EXIT_FAILURE);
                }
                times.actime = srcStatBuf.st_atime;
                times.modtime = srcStatBuf.st_mtime;
                if (utime(destName, &times) < 0) {
                        perror(destName);
-                       exit FALSE;
+                       exit(EXIT_FAILURE);
                }
        }
 
@@ -1713,11 +1713,21 @@ void xregcomp(regex_t *preg, const char *regex, int cflags)
                int errmsgsz = regerror(ret, preg, NULL, 0);
                char *errmsg = xmalloc(errmsgsz);
                regerror(ret, preg, errmsg, errmsgsz);
-               fatalError("bb_regcomp: %s\n", errmsg);
+               fatalError("xregcomp: %s\n", errmsg);
        }
 }
 #endif
 
+#if defined BB_UNIQ
+FILE *xfopen(const char *path, const char *mode)
+{
+       FILE *fp;
+       if ((fp = fopen(path, mode)) == NULL)
+               fatalError("%s: %s\n", path, strerror(errno));
+       return fp;
+}
+#endif
+
 /* END CODE */
 /*
 Local Variables: